Dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrid

Abstract

The invention discloses a dynamic anti-disturbance coordinated scheduling method for a hybrid energy micro-grid, and aims to solve the problems of poor stability, high cost, low consumption rate and the like of the hybrid energy micro-grid caused by disturbance such as renewable energy fluctuation and load abrupt change. The method comprises the following steps: firstly, constructing a system model containing multiple units, and collecting and preprocessing real-time operation data; establishing a multi-target scheduling objective function, and solving an initial scheduling instruction by adopting an improved predictive control algorithm; and finally, generating a coordinated scheduling strategy by combining the disturbance information dynamic correction instruction. Meanwhile, a security constraint and scheduling effect evaluation link is set, and mechanisms such as multi-time scale scheduling and layered response are adopted. The method can significantly improve the anti-disturbance capability and stability of the system, reduces the operation cost by 15%, improves the renewable energy consumption rate by more than 8%, is adaptive to multiple scenes, guarantees the safe operation, and is suitable for various hybrid energy microgrids.

Description

TECHNICAL FIELD

[0001] The application relates to a hybrid energy micro-grid, in particular to a hybrid energy micro-grid dynamic anti-disturbance coordinated scheduling method. BACKGROUND

[0002] Under the promotion of the double carbon target, the hybrid energy micro-grid can efficiently integrate photovoltaic power, wind power and other renewable energy, becomes the core carrier of energy transformation, and is widely used in industrial parks, rural areas, islands and other multiple scenes. Its core value lies in improving the utilization rate of renewable energy, but it faces significant stability challenges in operation: photovoltaic power and wind power output are strongly fluctuant due to natural factors such as light and wind speed, and are superimposed with random disturbances such as load mutation and equipment failure, which can easily cause system frequency deviation and voltage fluctuation, and even lead to power interruption. The existing scheduling method cannot solve the above problems, and has multiple key defects. The response lag problem is prominent, for example, the fixed cycle scheduling method disclosed in Chinese patent CN202310256789.1 uses a 1-hour scheduling cycle, which cannot respond to second-level load mutation, and experiments show that its frequency deviation recovery time is more than 15 seconds, and the hybrid energy storage system is not used, and the power imbalance rate is as high as 20%. The prediction error is also crucial. As shown in "Micro-grid scheduling research based on traditional MPC" (Power System Automation, 2022, Vol. 36), the traditional method does not introduce a disturbance observer, and the prediction error of renewable energy is more than 10%, which directly leads to an increase of 18% in operation cost. In addition, the traditional method has the problem of imbalance of multi-objective optimization, either ignoring the economy to cause high operation cost or not giving priority to renewable energy consumption to cause energy waste. At the same time, the scene adaptability is poor, it is difficult to adapt to micro-grids of different scales and grid-connected / off-grid operation modes, and there is a lack of comprehensive safety constraints, the risk of equipment over-limit operation is high, and the operation and maintenance pressure is increased. In summary, the existing scheduling method cannot balance the response speed, prediction accuracy, economy and adaptability, and it is urgent to build a more efficient coordinated scheduling technology, and the "layered response mechanism" and "improved prediction control algorithm" proposed in the application are the breakthrough solutions for the above problems. SUMMARY

[0003] The technical problem to be solved by the application is to overcome the defects of the above-mentioned technologies, and to provide a hybrid energy micro-grid dynamic anti-disturbance coordinated scheduling method.

[0004] To solve the above technical problems, the technical scheme provided by the application is a hybrid energy micro-grid dynamic anti-disturbance coordinated scheduling method, comprising the following steps:

[0005] (1) Construct a hybrid energy micro-grid system model, which at least includes renewable energy power generation units, energy storage units, controllable distributed power sources and load units, wherein each unit is connected to a common DC bus or AC bus through a power electronic converter, forming a multi-source collaborative power supply network; The system model needs to synchronously include the physical characteristic parameters, operating constraint conditions and interface characteristics of each unit, the physical characteristic parameters including the power-efficiency curve of the power generation unit, the charge-discharge rate characteristic of the energy storage unit, and the time sequence variation law of the load unit;

[0006] (2) Collect real-time operation data of the hybrid energy micro-grid system, including the output power of each power generation unit, the state of charge of the energy storage unit, the load demand and the system disturbance information, the data collection is realized through intelligent sensing terminals deployed at key nodes, the time stamp synchronization technology is adopted to ensure the spatio-temporal consistency of multi-source data, and the collected raw data needs to be preprocessed through filtering, outlier rejection and normalization to form a standardized data set meeting the input requirements of the scheduling algorithm;

[0007] (3) Based on the real-time operation data, a dynamic anti-disturbance scheduling objective function is established, which takes the suppression of power fluctuation, minimization of operation cost and improvement of energy utilization rate as the optimization target, and the construction of the objective function needs to consider the difference of the operation mode (grid-connected / off-grid) of the micro-grid, in the grid-connected mode, the interactive power characteristics with the main grid are optimized, and in the off-grid mode, the system frequency and voltage stability are prioritized;

[0008] (4) An improved predictive control algorithm is used to solve the objective function to obtain the initial scheduling instruction of each controllable unit, the solving process includes a constraint condition checking link, which ensures that the initial scheduling instruction meets the system safe operation boundary through a feasibility judgment mechanism, and the relaxation variable adjustment method is used to modify the infeasible solution;

[0009] (5) According to the system disturbance information, the initial scheduling instruction is dynamically modified to generate the final coordinated scheduling strategy and is sent to each execution unit, the dynamic modification process adopts a hierarchical response mechanism, the local hardware protection fast response is adopted for millisecond-level disturbance, the energy storage unit is used for real-time compensation for second-level disturbance, and the controllable power source output adjustment is used for minute-level disturbance, so as to realize accurate suppression of disturbances of different time scales.

[0010] As an improvement, the renewable energy power generation unit in step (1) includes a photovoltaic power generation module and a wind power generation module, the energy storage unit adopts a hybrid energy storage system composed of lithium batteries and super capacitors, and the controllable distributed power source includes a micro gas turbine and a diesel generator; The total output power of the hybrid energy storage system satisfies:

[0011] P ess =P bat +Psc

[0012] Among them, P ess P represents the total output power of the hybrid energy storage system. bat For the output power of the lithium battery, P sc The formula represents the output power of the supercapacitor. It characterizes the power distribution relationship between the two energy storage devices in a hybrid energy storage system, enabling coordinated control where short-term high-frequency disturbances are responded to by the supercapacitor, and long-term low-frequency fluctuations are compensated by the lithium battery. In this hybrid energy storage system, the time constant t used to separate the high-frequency and low-frequency components is an adaptive parameter. The energy management system monitors the state of charge of the lithium battery, the voltage of the supercapacitor, and the spectral characteristics of the microgrid power fluctuations in real time, dynamically adjusting the values ​​to achieve an optimal balance between system lifespan, efficiency, and cost. This division of labor reduces the number of charge-discharge cycles of the lithium battery, extending its lifespan.

[0013] As an improvement, the system disturbance information in step (2) includes renewable energy power prediction error, load change signal, communication network delay, and equipment fault information. The real-time operating data is collected through a distributed sensor network with a sampling frequency of not less than 1Hz. The formula for calculating the renewable energy power prediction error is:

[0014] ΔP re =P re,act -P re,pre

[0015] Wherein, ΔP re For renewable energy power prediction error, P re,act P represents the actual output power. re,pre This formula is used to predict output power and quantify the uncertainty of renewable energy output. It provides a basis for judging the disturbance amplitude for subsequent anti-disturbance dispatch. When the prediction error is positive, it means that the actual output is higher than the predicted value. The system needs to balance the output by charging energy storage or reducing the output of renewable energy. When the prediction error is negative, it means that the actual output is lower than the predicted value. The system needs to call the backup power or release energy storage for compensation.

[0016] As an improvement, the objective function in step (3) is constructed using a weighted summation method, comprehensively considering three types of indicators: power fluctuation mitigation, operating costs, and energy utilization rate. The expression is as follows:

[0017] minF=ω1F1+ω2F2+ω3F3

[0018] In this formula, F1 is the power fluctuation mitigation index, F2 is the system operating cost, and F3 is the energy utilization rate index. ω1, ω2, and ω3 are weight coefficients that satisfy ω1+ω2+ω3=1. This formula achieves collaborative decision-making for multi-objective optimization through weight allocation. The priority of each index can be dynamically adjusted according to the microgrid operation scenario. In emergency scenarios such as grid failures, the weight of ω1 is increased to above 0.6 to prioritize system stability. In normal economic operation scenarios, the weight of ω2 is set to 0.5-0.6 to highlight the cost optimization objective. During periods of high renewable energy generation, the weight of ω3 is increased to above 0.5 to improve the absorption rate.

[0019] As an improvement, the power fluctuation mitigation index F1 is calculated using the power fluctuation variance at the microgrid grid connection point, and its expression is:

[0020]

[0021] Among them, P grid,i Let be the grid connection power at time i. The average power at the grid connection point is denoted by n, and the number of sampling points is denoted by n. This formula is used to quantify the power fluctuation at the grid connection point. The smaller the value, the better the system power stability. In actual calculations, the value of n is determined according to the scheduling cycle. The intraday scheduling layer uses n = 96 (15-minute interval, 24 hours), and the real-time adjustment layer uses n = 60 (1-minute interval, 1 hour). Multi-scale fluctuation assessment is achieved through calculations of different window lengths.

[0022] As an improvement, the system operating cost F2 includes fuel cost, maintenance cost, and energy storage degradation cost, wherein the fuel cost is calculated using the following formula:

[0023]

[0024] Among them, a k b k c k Let P be the fuel cost coefficient for the k-th controllable power source. k,t Let Δt be the output at time t, m be the total number of controllable power sources, and Δt be the scheduling time interval. This formula is used to accurately calculate the fuel consumption cost of controllable power sources, providing a quantitative basis for economic optimization. The coefficient 'a' is used in this formula. k b k c k The quadratic coefficient b of the micro gas turbine is determined based on the equipment type. k The large absolute value reflects its nonlinear cost characteristics; the coefficient 'a' of the first term of the diesel generator... k It dominates, reflecting its approximately linear fuel consumption pattern.

[0025] As an improvement, the improved predictive control algorithm in step (4) is based on the model predictive control framework, introduces a disturbance observer to predict the fluctuations in renewable energy output in real time, and uses a rolling optimization window to achieve multi-timescale coordination. The optimization window includes an intraday scheduling layer (15-minute level) and a real-time adjustment layer (1-minute level); the coordination relationship between the two timescales satisfies:

[0026] T day =N×T real

[0027] Among them, T day For intraday scheduling cycles (15 minutes), T real The real-time adjustment period is 1 minute, where N is an integer and N = 15. This formula is used to define the time scale correlation between different scheduling levels, realizing the organic combination of global optimization and real-time response. The intraday scheduling layer generates the output plan curve for 24 hours based on ultra-short-term forecast data. The real-time adjustment layer corrects the intraday plan based on the latest measurement data in each 1-minute cycle. The correction range does not exceed ±20% of the intraday plan value to ensure the smoothness of the scheduling strategy.

[0028] As an improvement, the dynamic correction process in step (5) includes: when the system disturbance amplitude is detected to exceed a preset threshold, a fast response mechanism is triggered, the energy storage unit is preferentially called for power compensation, and the output plan of the controllable distributed power source is adjusted. The preset threshold is set to 5%-15% of the system rated power according to the microgrid capacity level; the disturbance amplitude judgment formula is:

[0029] |ΔP|≥α×P rated

[0030] Where |ΔP| is the absolute value of the disturbance power, α is the threshold coefficient (5%-15%), and P rated The system's rated power is used to determine whether a fast response mechanism is triggered, ensuring that disturbances are effectively suppressed before exceeding the system's tolerance range. For small microgrids with a capacity of less than 500 kVA, α is set to 10%-15% to improve response sensitivity. For medium and large microgrids with a capacity of more than 1000 kVA, α is set to 5%-10% to avoid increased operating losses caused by frequent responses.

[0031] As an improvement, safety constraints are also included, which include: upper and lower limits of output for each power generation unit, state of charge (SBC) constraints for energy storage units (20%-90%), system power balance constraints, and network power flow constraints; wherein the expression for the SBC constraint of the energy storage unit is:

[0032] SOC min ≤SOC t ≤SOC max

[0033] Among them, SOC t Let SOC be the state of charge of the stored energy at time t. min =20%, SOC max =90%; This formula is used to prevent overcharging and over-discharging of energy storage units, extend equipment life and ensure dispatch flexibility. In actual operation, when a period of high renewable energy generation is predicted in the next 24 hours, the SOC will be adjusted to below 30% in advance to reserve charging space. When a peak load or severe weather is predicted to cause a sharp decrease in renewable energy output, the SOC will be increased to above 80% in advance to reserve emergency power.

[0034] As an improvement, a scheduling effect evaluation step is also included: by comparing the changes in system frequency deviation, voltage fluctuation range, and economic indicators before and after the disturbance, the effectiveness of the dynamic disturbance-resistant coordinated scheduling method is evaluated. When the evaluation indicators do not meet the preset requirements, the algorithm parameters are automatically updated and the scheduling process is re-executed; the system frequency deviation calculation formula is:

[0035] Δf=|f t -f nominal |

[0036] Where Δf is the absolute value of the frequency deviation, f t Let f be the system frequency at time t. nominal The rated frequency is 50Hz or 60Hz. This formula is used to quantify the frequency stability of the system and is one of the core indicators for evaluating the anti-disturbance effect. The evaluation standard is: within 3 seconds after the disturbance occurs, Δf should be controlled within ±0.2Hz (grid-connected mode) or ±0.5Hz (off-grid mode), and recover to within ±0.1Hz within 10 seconds. When the standard is not met for 3 consecutive evaluations, the algorithm parameter optimization process is automatically started, and the predictive control window length and weight coefficients are readjusted.

[0037] The advantages of this invention compared with the prior art are as follows: This technical solution can significantly improve the anti-disturbance capability and operational stability of hybrid energy microgrids. Through a multi-timescale scheduling framework and hierarchical response mechanism, it can quickly respond to disturbances of different durations, control frequency deviation within 3 seconds, restore stability within 10 seconds, reduce the risk of power outages, and quantify and predict errors in advance to prevent and control them, thereby significantly reducing the power fluctuation variance at the grid connection point.

[0038] At the same time, it significantly reduces operating costs by optimizing the objective function through multiple objectives, adjusting the cost optimization weights according to the scenario, calculating fuel costs in detail and optimizing equipment output, and reducing energy storage and controllable power losses by combining constraints, thereby extending equipment life.

[0039] It can also improve the renewable energy absorption rate, prioritize absorption during peak power generation, and use the "prediction-storage-dispatch" mechanism to buffer excess power through energy storage and adjust equipment status in real time, thereby increasing the absorption rate by more than 8%, reducing energy waste, and contributing to low-carbon development.

[0040] In addition, it enhances scenario adaptability, adjusts response thresholds according to microgrid scale, adapts to grid-connected and off-grid mode requirements, modular design facilitates the addition of new equipment and reduces upgrade costs, and also builds a full-cycle safety guarantee system to reduce equipment failures, alleviate operation and maintenance pressure, and provide support for the large-scale application of microgrids. Attached Figure Description

[0041] Figure 1 This is a flowchart illustrating the overall method of the dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrids according to the present invention.

[0042] Figure 2 This is a flowchart of the dynamic disturbance mitigation correction process for the dynamic disturbance mitigation coordination scheduling method for hybrid energy microgrids of the present invention. Detailed Implementation

[0043] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings, which illustrate embodiments of the present application. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that the disclosure of this application will be thorough and complete.

[0044] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application.

[0045] It is understood that spatial relation terms such as "below," "under," "below," "below," "above," "over," etc., can be used here to describe the relationship between one element or feature shown in the figure and other elements or features. It should be understood that, in addition to the orientation shown in the figure, spatial relation terms also include different orientations of the device in use and operation. For example, if the device in the figure is flipped, the element or feature described as "below" or "under" or "below" of the other element or feature will be oriented "over" the other element or feature. Therefore, the exemplary terms "below" and "under" can include both upper and lower orientations. Furthermore, the device may also include other orientations, such as being rotated 90 degrees or other orientations, and the spatial descriptive terms used herein will be interpreted accordingly.

[0046] It should be noted that when one element is considered to be "connected" to another element, it can be directly connected to the other element or connected to the other element through an intermediary element. In the following embodiments, "connection" should be understood as "electrical connection," "communication connection," etc., if the connected circuits, modules, units, etc., have the transmission of electrical signals or data between them.

[0047] When used herein, the singular forms of “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “comprising,” “including,” or “having,” etc., specify the presence of the stated feature, whole, step, operation, component, part, or combination thereof, but do not preclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof.

[0048] Referring to the attached figures, the dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrids includes: (1) constructing a hybrid energy microgrid system model, wherein the hybrid energy microgrid system includes at least renewable energy generation units, energy storage units, controllable distributed power sources and load units, wherein each unit is connected to a common DC bus or AC bus through a power electronic converter to form a multi-source coordinated power supply network; the system model needs to synchronously include the physical characteristic parameters, operating constraints and interface characteristics of each unit, wherein the physical characteristic parameters include the power-efficiency curve of the generation unit, the charge-discharge rate characteristics of the energy storage unit, and the time-series variation law of the load unit;

[0049] (2) Collect real-time operating data of the hybrid energy microgrid system. The real-time operating data includes the output power of each power generation unit, the state of charge of the energy storage unit, the load demand and system disturbance information. Data collection is achieved through intelligent sensing terminals deployed at each key node. Timestamp synchronization technology is used to ensure the spatiotemporal consistency of multi-source data. The collected raw data needs to be filtered, outlier removal and normalization preprocessing to form a standardized dataset that meets the input requirements of the scheduling algorithm.

[0050] (3) Establish a dynamic anti-disturbance scheduling objective function based on real-time operation data. The objective function aims to smooth power fluctuations, minimize operating costs and improve energy utilization. The construction of the objective function needs to consider the differences in the operation mode (grid-connected / off-grid) of the microgrid. In the grid-connected mode, the focus is on optimizing the interactive power characteristics with the main grid. In the off-grid mode, priority is given to ensuring the stability of system frequency and voltage.

[0051] (4) An improved predictive control algorithm is used to solve the objective function to obtain the initial scheduling instructions for each controllable unit. The solution process includes a constraint verification step. The feasibility judgment mechanism ensures that the initial scheduling instructions meet the system's safe operation boundary. For infeasible solutions, the relaxation variable adjustment method is used for correction.

[0052] (5) The initial scheduling instructions are dynamically corrected based on the system disturbance information, the final coordinated scheduling strategy is generated and sent to each execution unit. The dynamic correction process adopts a hierarchical response mechanism. Local hardware protection is used for rapid response to millisecond-level disturbances, energy storage units are used for real-time compensation for second-level disturbances, and controllable power output is used for minute-level disturbances, so as to achieve precise suppression of disturbances at different time scales.

[0053] As an improvement, the renewable energy power generation unit in step (1) includes a photovoltaic power generation module and a wind power generation module; the energy storage unit adopts a hybrid energy storage system composed of lithium batteries and supercapacitors; and the controllable distributed power source includes a micro gas turbine and a diesel generator. The total output power of the hybrid energy storage system satisfies:

[0054] P ess =P bat +P sc

[0055] Among them, P ess P represents the total output power of the hybrid energy storage system. bat For the output power of the lithium battery, P sc The formula represents the output power of the supercapacitor. It characterizes the power distribution relationship between the two energy storage devices in a hybrid energy storage system, enabling coordinated control where short-term high-frequency disturbances are responded to by the supercapacitor, and long-term low-frequency fluctuations are compensated by the lithium battery. In this hybrid energy storage system, the time constant t used to separate the high-frequency and low-frequency components is an adaptive parameter. The energy management system monitors the state of charge of the lithium battery, the voltage of the supercapacitor, and the spectral characteristics of the microgrid power fluctuations in real time, dynamically adjusting the values ​​to achieve an optimal balance between system lifespan, efficiency, and cost. This division of labor reduces the number of charge-discharge cycles of the lithium battery, extending its lifespan.

[0056] As an improvement, the system disturbance information in step (2) includes renewable energy power prediction error, load change signal, communication network delay, and equipment fault information. The real-time operating data is collected through a distributed sensor network with a sampling frequency of not less than 1Hz. The formula for calculating the renewable energy power prediction error is:

[0057] ΔP re =P re,act -P re,pre

[0058] Wherein, ΔP re For renewable energy power prediction error, P re,act P represents the actual output power. re,preThis formula is used to predict output power and quantify the uncertainty of renewable energy output. It provides a basis for judging the disturbance amplitude for subsequent anti-disturbance dispatch. When the prediction error is positive, it means that the actual output is higher than the predicted value. The system needs to balance the output by charging energy storage or reducing the output of renewable energy. When the prediction error is negative, it means that the actual output is lower than the predicted value. The system needs to call the backup power or release energy storage for compensation.

[0059] As an improvement, the objective function in step (3) is constructed using a weighted summation method, comprehensively considering three types of indicators: power fluctuation mitigation, operating costs, and energy utilization rate. The expression is as follows:

[0060] minF=ω1F1+ω2F2+ω3F3

[0061] In this formula, F1 is the power fluctuation mitigation index, F2 is the system operating cost, and F3 is the energy utilization rate index. ω1, ω2, and ω3 are weight coefficients that satisfy ω1+ω2+ω3=1. This formula achieves collaborative decision-making for multi-objective optimization through weight allocation. The priority of each index can be dynamically adjusted according to the microgrid operation scenario. In emergency scenarios such as grid failures, the weight of ω1 is increased to above 0.6 to prioritize system stability. In normal economic operation scenarios, the weight of ω2 is set to 0.5-0.6 to highlight the cost optimization objective. During periods of high renewable energy generation, the weight of ω3 is increased to above 0.5 to improve the absorption rate.

[0062] As an improvement, the power fluctuation mitigation index F1 is calculated using the power fluctuation variance at the microgrid grid connection point, and its expression is:

[0063]

[0064] Among them, P grid,i Let be the grid connection power at time i. The average power at the grid connection point is denoted by n, and the number of sampling points is denoted by n. This formula is used to quantify the power fluctuation at the grid connection point. The smaller the value, the better the system power stability. In actual calculations, the value of n is determined according to the scheduling cycle. The intraday scheduling layer uses n = 96 (15-minute interval, 24 hours), and the real-time adjustment layer uses n = 60 (1-minute interval, 1 hour). Multi-scale fluctuation assessment is achieved through calculations of different window lengths.

[0065] As an improvement, the system operating cost F2 includes fuel cost, maintenance cost, and energy storage degradation cost, wherein the fuel cost is calculated using the following formula:

[0066]

[0067] Among them, a k b k c k Let P be the fuel cost coefficient for the k-th controllable power source.k,t Let Δt be the output at time t, m be the total number of controllable power sources, and Δt be the scheduling time interval. This formula is used to accurately calculate the fuel consumption cost of controllable power sources, providing a quantitative basis for economic optimization. The coefficient 'a' is used in this formula. k b k c k The quadratic coefficient b of the micro gas turbine is determined based on the equipment type. k The large absolute value reflects its nonlinear cost characteristics; the coefficient 'a' of the first term of the diesel generator... k It dominates, reflecting its approximately linear fuel consumption pattern.

[0068] As an improvement, the improved predictive control algorithm in step (4) is based on the model predictive control framework, introduces a disturbance observer to predict the fluctuations in renewable energy output in real time, and uses a rolling optimization window to achieve multi-timescale coordination. The optimization window includes an intraday scheduling layer (15-minute level) and a real-time adjustment layer (1-minute level); the coordination relationship between the two timescales satisfies:

[0069] T day =N×T real

[0070] Among them, T day For intraday scheduling cycles (15 minutes), T real The real-time adjustment period is 1 minute, where N is an integer and N = 15. This formula is used to define the time scale correlation between different scheduling levels, realizing the organic combination of global optimization and real-time response. The intraday scheduling layer generates the output plan curve for 24 hours based on ultra-short-term forecast data. The real-time adjustment layer corrects the intraday plan based on the latest measurement data in each 1-minute cycle. The correction range does not exceed ±20% of the intraday plan value to ensure the smoothness of the scheduling strategy.

[0071] As an improvement, the dynamic correction process in step (5) includes: when the system disturbance amplitude is detected to exceed a preset threshold, a fast response mechanism is triggered, the energy storage unit is preferentially called for power compensation, and the output plan of the controllable distributed power source is adjusted. The preset threshold is set to 5%-15% of the system rated power according to the microgrid capacity level; the disturbance amplitude judgment formula is:

[0072] |ΔP|≥α×P rated

[0073] Where |ΔP| is the absolute value of the disturbance power, α is the threshold coefficient (5%-15%), and P ratedThe system's rated power is used to determine whether a fast response mechanism is triggered, ensuring that disturbances are effectively suppressed before exceeding the system's tolerance range. For small microgrids with a capacity of less than 500 kVA, α is set to 10%-15% to improve response sensitivity. For medium and large microgrids with a capacity of more than 1000 kVA, α is set to 5%-10% to avoid increased operating losses caused by frequent responses.

[0074] As an improvement, safety constraints are also included, which include: upper and lower limits of output for each power generation unit, state of charge (SBC) constraints for energy storage units (20%-90%), system power balance constraints, and network power flow constraints; wherein the expression for the SBC constraint of the energy storage unit is:

[0075] SOC min ≤SOC t ≤SOC max

[0076] Among them, SOC t Let SOC be the state of charge of the stored energy at time t. min =20%, SOC max =90%; This formula is used to prevent overcharging and over-discharging of energy storage units, extend equipment life and ensure dispatch flexibility. In actual operation, when a period of high renewable energy generation is predicted in the next 24 hours, the SOC will be adjusted to below 30% in advance to reserve charging space. When a peak load or severe weather is predicted to cause a sharp decrease in renewable energy output, the SOC will be increased to above 80% in advance to reserve emergency power.

[0077] As an improvement, a scheduling effect evaluation step is also included: by comparing the changes in system frequency deviation, voltage fluctuation range, and economic indicators before and after the disturbance, the effectiveness of the dynamic disturbance-resistant coordinated scheduling method is evaluated. When the evaluation indicators do not meet the preset requirements, the algorithm parameters are automatically updated and the scheduling process is re-executed; the system frequency deviation calculation formula is:

[0078] Δf=|f t -f nominal |

[0079] Where Δf is the absolute value of the frequency deviation, f t Let f be the system frequency at time t. nominal The rated frequency is 50Hz or 60Hz. This formula is used to quantify the frequency stability of the system and is one of the core indicators for evaluating the anti-disturbance effect. The evaluation standard is: within 3 seconds after the disturbance occurs, Δf should be controlled within ±0.2Hz (grid-connected mode) or ±0.5Hz (off-grid mode), and recover to within ±0.1Hz within 10 seconds. When the standard is not met for 3 consecutive evaluations, the algorithm parameter optimization process is automatically started, and the predictive control window length and weight coefficients are readjusted.

[0080] This invention discloses a dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrids, which aims to solve the problems of poor system stability and low operating efficiency caused by disturbances such as renewable energy output fluctuations and load changes in hybrid energy microgrids. By constructing a multi-timescale coordinated scheduling framework, it achieves dynamic suppression of various disturbances, while taking into account both system economy and energy utilization.

[0081] I. Model Construction of Hybrid Energy Microgrid System:

[0082] The hybrid energy microgrid system model construction steps (step 1) of this invention specifically include the following:

[0083] The system model must fully encompass four core components: renewable energy generation units, energy storage units, controllable distributed power sources, and load units. Each unit is connected to a common DC or AC bus via power electronic converters (such as inverters and rectifiers) to form a multi-source collaborative power supply network. The renewable energy generation units include photovoltaic (PV) power generation modules and wind power generation modules. PV power generation modules use monocrystalline silicon photovoltaic modules, whose output characteristics are significantly affected by light intensity and ambient temperature; the model must include light-temperature-power characteristic curves. Wind power generation modules use horizontal axis wind turbines; their output power is approximately proportional to the cube of the wind speed; the model must include cut-in wind speed, rated wind speed, and cut-out wind speed parameters to define their operating range.

[0084] The energy storage unit employs a hybrid energy storage system composed of lithium batteries and supercapacitors. The lithium batteries are lithium iron phosphate batteries, characterized by high energy density (150-200Wh / kg) and long cycle life (over 2000 cycles), primarily used to smooth out long-term low-frequency power fluctuations. The supercapacitors feature high power density (1000-5000W / kg) and fast charge / discharge response (millisecond level), mainly used to absorb short-term high-frequency disturbances. Both are connected in parallel to the DC bus via a bidirectional DC / DC converter, and the total output power meets the following requirements:

[0085] P ess =P bat +P sc

[0086] Among them, P ess P represents the total output power of the hybrid energy storage system. bat For the output power of the lithium battery, P scThis formula defines the power distribution relationship between the two energy storage devices. In actual operation, the total power command is decomposed into a low-frequency component (handled by the lithium battery) and a high-frequency component (handled by the supercapacitor) through a Butterworth filter. The cutoff frequency is usually set to 0.01Hz, so that the supercapacitor is responsible for responding to power fluctuations within 10 seconds, and the lithium battery is responsible for handling power deficits of 10 seconds to 10 minutes. This reduces the number of charge-discharge cycles of the lithium battery and extends its service life.

[0087] Controllable distributed power sources include micro gas turbines and diesel generators, serving as backup and peak-shaving power sources for the system. Micro gas turbines use natural gas as fuel, with a power generation efficiency of 25%-35%, and are characterized by fast start-up and shutdown (5-10 minutes) and low pollution emissions. Diesel generators have a power generation efficiency of 30%-40%, but have a longer start-up time (10-15 minutes) and are mainly used as emergency power sources in emergency situations.

[0088] The load units include residential load, commercial load and industrial load. The model needs to reflect the time-series variation of different types of loads. For example, residential load has morning peak (7:00-9:00) and evening peak (18:00-22:00) characteristics, while industrial load maintains a relatively stable base load characteristic.

[0089] When constructing the system model, the physical characteristic parameters, operating constraints, and interface characteristics of each unit need to be entered simultaneously. The physical characteristic parameters include the power-efficiency curve of the power generation unit (such as the quadratic curve of the efficiency of a micro gas turbine as a function of output), the charge-discharge rate characteristics of the energy storage unit (such as the maximum charge-discharge rate of 1C for lithium batteries and 5C for supercapacitors), and the time-series variation pattern of the load unit (a 24-hour load curve fitted from historical data).

[0090] II. Real-time Data Acquisition and Processing:

[0091] The real-time data acquisition and processing process described in step (2) is implemented as follows:

[0092] Data acquisition is achieved through intelligent sensing terminals deployed at key nodes, including DC power sensors installed at the output of the photovoltaic array, AC power sensors at the wind turbine outlet, voltage / current sensors in the energy storage unit, smart meters on the load side, and a synchronization phasor measurement unit (PMU) at the system grid connection point. All sensing terminals use the IEEE 1588 Precise Time Protocol (PTP) to achieve timestamp synchronization, ensuring the spatiotemporal consistency of multi-source data, with synchronization accuracy controlled within 100 microseconds.

[0093] The collected real-time operational data includes: output power of each power generation unit (such as real-time output of photovoltaic power, real-time output of wind power, and real-time output of micro gas turbines), state of charge (SOC) of energy storage units, load demand (real-time load values ​​statistically categorized by type), and system disturbance information. Specifically, the system disturbance information includes: renewable energy power prediction errors (deviations between predicted and actual values ​​for photovoltaic and wind power), load surge signals (events where the load change rate exceeds 5% / minute), communication network latency (data transmission delays exceeding 100 milliseconds), and equipment fault information (such as alarm signals for inverter overcurrent and generator overspeed).

[0094] The data sampling frequency is no less than 1Hz, meaning data is collected at least once per second. For rapidly changing disturbance signals (such as sudden load changes), the sampling frequency can be increased to 10Hz to ensure that the complete waveform of the disturbance is captured. The collected raw data needs to undergo a preprocessing process: first, a moving average filtering method is used to remove high-frequency noise, with the window size set to 5 sampling points; then, outliers (data that deviate from the average value by more than 3 times the standard deviation) are identified and removed using the 3σ criterion; finally, normalization processing is performed to convert parameters of different dimensions to the [0,1] interval, forming a standardized dataset that meets the input requirements of the scheduling algorithm.

[0095] The renewable energy power forecast error is calculated using the following formula:

[0096] ΔP re =P re,act -P re,pre

[0097] Wherein, ΔP re For renewable energy power prediction error, P re,act P represents the actual output power. re,pre This is the ultra-short-term predicted output power based on an LSTM neural network (prediction duration of 1 hour, rolling update cycle of 15 minutes). This formula quantifies the uncertainty of renewable energy output. When the prediction error is positive, it indicates that the actual output is higher than the predicted value, and the system needs to balance power by charging energy storage or reducing renewable energy output (where wind and solar curtailment is permitted). When the prediction error is negative, it indicates that the actual output is lower than the predicted value, and the system needs to call upon controllable distributed power sources or release energy storage for power compensation. Error data will be transmitted to the disturbance observer in real time, providing accurate disturbance amplitude judgment for subsequent anti-disturbance scheduling.

[0098] III. Establishment of the Objective Function for Dynamic Disturbance-Resistant Scheduling:

[0099] The process of establishing the dynamic disturbance-resistant scheduling objective function in step (3) is as follows:

[0100] The objective function is constructed using a weighted summation method, with three main optimization objectives: smoothing power fluctuations, minimizing operating costs, and improving energy utilization. This method adapts to the needs of different operating scenarios. The objective function expression is as follows:

[0101] minF=ω1F1+ω2F2+ω3F3

[0102] Wherein, F1 is the power fluctuation mitigation index, F2 is the system operating cost, F3 is the energy utilization rate index, and ω1, ω2, and ω3 are weighting coefficients that satisfy ω1+ω2+ω3=1.

[0103] This formula achieves collaborative decision-making for multi-objective optimization through weight allocation, and can dynamically adjust the priority of each indicator according to the microgrid operation scenario. Specifically, in emergency scenarios such as grid failures and large disturbances, system stability is crucial, and the weight of ω1 is increased to above 0.6 to prioritize the smoothing effect of power fluctuations; in normal economic operation scenarios, the weight of ω2 is set to 0.5-0.6 to highlight the cost optimization objective; during periods of high renewable energy generation (such as when there is strong sunlight at noon or strong winds at night), the weight of ω3 is increased to above 0.5 to improve the renewable energy absorption rate and reduce wind and solar curtailment.

[0104] The construction of the objective function needs to consider the differences in the microgrid's operating mode (grid-connected / off-grid). In grid-connected mode, the objective function focuses on optimizing the interaction power characteristics with the main grid to minimize power fluctuations at the grid connection point and reduce the impact on the main grid. In off-grid mode, the objective function prioritizes ensuring system frequency and voltage stability, and in this case, penalty terms for frequency deviation and voltage deviation will be added to the F1 index.

[0105] The power fluctuation mitigation index F1 is calculated using the power fluctuation variance at the microgrid grid connection point, and its expression is:

[0106]

[0107] Among them, P grid,i Let be the grid connection power at time i (positive values ​​indicate electricity purchased from the main grid, negative values ​​indicate electricity sold to the main grid). Let n be the average power at the grid connection point and n be the number of sampling points. This formula is used to quantify the degree of power fluctuation at the grid connection point; the smaller the value, the better the system power stability. In actual calculations, the value of n is determined according to the scheduling cycle: the intraday scheduling layer (15-minute level) uses n = 96 (15-minute intervals, covering 24 hours) to assess the power fluctuation trend throughout the day; the real-time adjustment layer (1-minute level) uses n = 60 (1-minute intervals, covering 1 hour) to monitor short-term power fluctuations. By calculating different window lengths, multi-scale fluctuation assessment is achieved, ensuring that the objective function reflects both global stability and real-time disturbances.

[0108] The system operating cost F2 includes fuel cost, maintenance cost, and energy storage degradation cost. Fuel cost is the most significant component, calculated using the following formula:

[0109]

[0110] Among them, a k b k c k P is the fuel cost coefficient for the k-th controllable power source (obtained by fitting the fuel consumption curve provided by the manufacturer). k,t Let t represent the output at time t, m represent the total number of controllable power sources, and Δt represent the scheduling time interval (15 minutes, or 900 seconds, for the intraday scheduling layer; and 1 minute, or 60 seconds, for the real-time adjustment layer). This formula is used to accurately calculate the fuel consumption cost of controllable power sources, providing a quantitative basis for economic optimization. Different types of controllable power sources have different cost coefficient characteristics: the quadratic coefficient b of a micro gas turbine... k The large absolute value reflects its nonlinear cost characteristics (efficiency varies significantly with output); the primary coefficient 'a' of the diesel generator. k It dominates, reflecting its approximately linear fuel consumption pattern.

[0111] Maintenance costs are calculated based on a fixed percentage of the controllable power output (e.g., 0.02 yuan / kWh for micro gas turbines and 0.015 yuan / kWh for diesel generators); energy storage degradation costs are related to the depth of charge and discharge and the number of cycles, and are estimated using a fixed coefficient of 0.05 yuan / kWh.

[0112] The energy utilization rate index F3 is a weighted value of renewable energy absorption rate and energy storage cycle efficiency. The renewable energy absorption rate is defined as the ratio of the actual renewable energy power absorbed to the total power generation. The energy storage cycle efficiency is the ratio of discharge to charging (approximately 90% for lithium batteries and approximately 95% for supercapacitors).

[0113] IV. Solving the Improved Predictive Control Algorithm:

[0114] The process of solving the objective function using the improved predictive control algorithm in step (4) is as follows:

[0115] The improved predictive control algorithm is based on the model predictive control (MPC) framework. Its core idea is to solve an optimization problem in the prediction time domain within each control cycle, based on the current system state and future disturbance predictions, to obtain a set of control sequences. Only the first control command is executed, and the process is repeated in the next cycle. This invention introduces a disturbance observer and a multi-timescale rolling optimization window into the traditional MPC to improve its adaptability to disturbances.

[0116] The disturbance observer employs the Extended Kalman Filter (EKF) algorithm to predict disturbances such as renewable energy output fluctuations and load changes in real time, based on historical data and real-time measurements. The prediction time domain matches the scheduling cycle (24 hours for intraday scheduling and 1 hour for real-time adjustment). The observer updates the prediction model parameters every 15 minutes to adapt to changes in weather and load patterns.

[0117] The multi-timescale rolling optimization window includes an intraday scheduling layer (15-minute level) and a real-time adjustment layer (1-minute level). The coordination relationship between the two timescales satisfies:

[0118] T day =N×T real

[0119] Among them, T day For intraday scheduling cycles (15 minutes), T real The real-time adjustment period is 1 minute, where N is an integer and N = 15. This formula defines the time scale correlation between different scheduling levels, achieving an organic combination of global optimization and real-time response.

[0120] The intraday scheduling layer generates output plan curves for each controllable unit within the next 24 hours based on ultra-short-term forecast data (solar power output forecast, wind power output forecast, and load forecast). The optimization objectives are primarily economic efficiency and energy utilization (ω2 and ω3 have higher weights). During the solution process, the particle swarm optimization (PSO) algorithm is used to optimize the objective function, with a population size of 50 and a maximum number of iterations of 100, to achieve a balance between computational efficiency and optimization accuracy.

[0121] The real-time adjustment layer revises the daily schedule based on the latest measurement data (including real-time output, SOC, and disturbance error) within each 1-minute cycle, with the optimization objective primarily focused on smoothing power fluctuations (ω1 has a higher weight). The revision process employs a quadratic programming algorithm, with computation time controlled within 500 milliseconds to ensure real-time performance. To guarantee the smoothness of the scheduling strategy, the real-time adjustment layer's revision of the daily schedule does not exceed ±20%, avoiding increased losses caused by frequent and significant equipment adjustments.

[0122] The solution process includes a constraint verification step, using a feasibility judgment mechanism to ensure that the initial scheduling instructions meet the system's safe operation boundaries. For infeasible solutions (such as those exceeding the equipment output limit or SOC constraints), a relaxation variable adjustment method is used for correction. This involves adding a penalty term to the objective function to make the optimization result as close as possible to the constraint boundaries, while recording the relaxation amount as a basis for subsequent scheduling adjustments.

[0123] V. Dynamic Correction and Coordination Scheduling Strategy Generation:

[0124] The process of dynamically correcting the initial scheduling command based on system disturbance information in step (5) is as follows:

[0125] The dynamic correction process employs a hierarchical response mechanism, using differentiated handling strategies for disturbances at different time scales: for millisecond-level disturbances (such as switching transients of power electronic equipment), local hardware protection (such as fast-acting fuses and current limiters) provides a rapid response; for second-level disturbances (such as sudden load changes and gusts of wind), energy storage units provide real-time power compensation; and for minute-level disturbances (such as a slow decline in photovoltaic output and a gradual increase in load gradient), controllable distributed power generation output adjustment is employed. This hierarchical response mechanism enables precise suppression of disturbances at different time scales, improving the system's disturbance immunity.

[0126] When the detected system disturbance amplitude exceeds a preset threshold, a rapid response mechanism is automatically triggered. The formula for determining the disturbance amplitude is:

[0127] |ΔP|≥α×P rated

[0128] Where |ΔP| is the absolute value of the disturbance power (calculated from the power change measured in real time), α is the threshold coefficient (5%-15%), and P rated This is the system's rated power (determined based on the microgrid's installed capacity).

[0129] This formula is used to determine whether a fast response mechanism is triggered, ensuring that disturbances are effectively suppressed before exceeding the system's tolerance range. The value of the threshold coefficient α is related to the microgrid capacity: for small microgrids with a capacity of less than 500 kVA, the system inertia is small and the disturbance resistance is weak, so α is set at 10%-15% to improve response sensitivity; for medium and large microgrids with a capacity of more than 1000 kVA, the system inertia is large, so α is set at 5%-10% to avoid increased operating losses caused by frequent responses.

[0130] Once the rapid response mechanism is activated, the energy storage units are prioritized for power compensation: the supercapacitor responds within 0.1 seconds, providing short-term high-power support; the lithium battery begins output within 1 second, providing continuous power compensation. Simultaneously, the system sends output adjustment commands to the controllable distributed power sources, the micro gas turbine completes output adjustment within 5 minutes, and the diesel generator starts as a backup within 10 minutes (only activated when energy storage capacity is insufficient).

[0131] The final coordination and scheduling strategy is issued through a collaborative architecture of edge computing and cloud computing: scheduling instructions with high real-time requirements (such as energy storage charging and discharging control and rapid load shedding) are processed and issued locally by edge nodes (industrial control computers installed locally in the microgrid), with transmission latency controlled within 100 milliseconds; global optimization instructions (such as daily output plans and equipment start-up and shutdown plans) are calculated by the cloud platform (servers deployed in remote data centers) and pushed to each edge node, with an update cycle of 15 minutes.

[0132] VI. Safety Constraints:

[0133] The method described in this invention also includes establishing security constraints to ensure the security and feasibility of the scheduling strategy, specifically including:

[0134] Output limits for each power generation unit: The output of each controllable power source shall not exceed its rated power, nor shall it be lower than the minimum technical output (e.g., the minimum output of a micro gas turbine is 30% of its rated power, and that of a diesel generator is 40%).

[0135] Energy storage unit state of charge constraints:

[0136] SOC min ≤SOC t ≤SOC max

[0137] Among them, SOC t Let SOC be the state of charge of the stored energy at time t. min =20%, SOC max =90%. This formula is used to prevent overcharging and over-discharging of energy storage units, extend equipment lifespan, and ensure dispatch flexibility. In actual operation, the system will adjust the SOC in advance based on forecast information: when a period of high renewable energy generation is predicted in the next 24 hours (such as midday on a sunny day), the SOC will be adjusted to below 30% in advance to reserve sufficient charging space; when a peak load or severe weather (such as continuous rain) is predicted to cause a sharp decrease in renewable energy output, the SOC will be increased to above 80% in advance to reserve emergency power.

[0138] System power balance constraint: At any given time, the total output of all power generation units (including renewable energy, controllable power sources, and energy storage discharge) must be equal to the total load demand (including user load, energy storage charging, and network losses).

[0139] Network power flow constraints: The transmission power of each line must not exceed its thermal stability limit, and the node voltage must be maintained within ±5% of the rated voltage (361-399V for a 380V system).

[0140] VII. Evaluation of Scheduling Effectiveness:

[0141] The method described in this invention also includes a scheduling effect evaluation step, which assesses the effectiveness of the dynamic anti-disturbance coordinated scheduling method by comparing changes in key system indicators before and after the disturbance. Evaluation indicators include system frequency deviation, voltage fluctuation range, and economic indicators (operating costs, renewable energy absorption rate).

[0142] The formula for calculating system frequency deviation is:

[0143] Δf=|f t -f nominal |

[0144] Where Δf is the absolute value of the frequency deviation, f t Let f be the system frequency at time t (measured in real time by PMU). nominal This is the rated frequency (50Hz in my country). This formula is used to quantify the frequency stability of a system and is one of the core indicators for evaluating its anti-disturbance performance.

[0145] The evaluation criteria are set according to the operating mode: In grid-connected mode, due to the supporting role of the main grid, it is required that Δf be controlled within ±0.2Hz within 3 seconds after the disturbance occurs, and recover to within ±0.1Hz within 10 seconds; In off-grid mode, the system inertia is smaller, and a slightly larger frequency deviation is allowed, requiring it to be controlled within ±0.5Hz within 3 seconds, and recover to within ±0.2Hz within 10 seconds.

[0146] Voltage fluctuation range is assessed using the maximum value of node voltage deviation, which must not exceed ±7% of the rated voltage. Economic indicators include: the percentage reduction in operating costs compared to traditional dispatching methods (target ≥10%) and the percentage increase in renewable energy integration rate (target ≥5%).

[0147] When the preset criteria are not met in three consecutive evaluations, the system automatically starts the algorithm parameter optimization process, readjusts the prediction control window length (e.g., shortens the real-time adjustment window from 1 hour to 30 minutes) and weight coefficients (e.g., increases the weight of ω1), and re-executes the scheduling process until the evaluation indicators meet the requirements.

[0148] Example effect verification:

[0149] To verify the effectiveness of the present invention, a 30-day field test was conducted on a hybrid energy microgrid (including 1MW photovoltaic, 500kW wind power, 200kWh lithium battery, 50kWh supercapacitor, 1 500kW micro gas turbine, and 1 300kW diesel generator) in an industrial park.

[0150] The experimental results show that after adopting the method of the present invention, the power fluctuation variance (F1) at the microgrid grid connection point is reduced by 42% compared with the traditional method; the average daily operating cost of the system (F2) is reduced by 15%; the renewable energy absorption rate (a component of F3) is increased by 8%; when encountering a 100kW load surge (approximately 10% of the system's rated power), the maximum frequency deviation is controlled within 0.3Hz, and the recovery time is only 5 seconds, which is far superior to the 0.8Hz and 15 seconds of the traditional method.

[0151] The above test results fully demonstrate that the dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrids described in this invention can effectively suppress power fluctuations, improve system stability, reduce operating costs, and enhance energy utilization, thus possessing significant technical advantages and application value.

[0152] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0153] The dynamic disturbance-resistant coordinated scheduling method for hybrid energy microgrids described in this invention, through innovative system modeling, data processing, algorithm design, and strategy optimization, breaks through the limitations of traditional scheduling methods from multiple dimensions such as technology, economy, and energy utilization. It provides a comprehensive solution for the stable, efficient, and economical operation of hybrid energy microgrids, and its beneficial effects are specifically reflected in the following aspects:

[0154] Significantly improving the disturbance immunity and operational stability of hybrid energy microgrids: In hybrid energy microgrids, the output of renewable energy sources (such as photovoltaic and wind power) is highly volatile due to natural conditions, and disturbances such as load abrupt changes and equipment failures occur frequently. These factors can easily lead to system frequency shifts, voltage fluctuations, and even power outages. This method fundamentally solves this core problem by constructing a "layered response + dynamic correction" disturbance immunity mechanism.

[0155] On the one hand, the multi-timescale coordinated scheduling framework designed in the method (15-minute level intraday scheduling layer and 1-minute level real-time adjustment layer) can achieve precise matching response to different types of disturbances: for millisecond-level transient disturbances of power electronic equipment switching, the fault source is quickly cut off by relying on local hardware protection to avoid the spread of disturbances; for second-level short-term disturbances such as gusts and load changes, millisecond-level response is achieved through a hybrid energy storage system (lithium battery + supercapacitor), in which the supercapacitor preferentially absorbs high-frequency fluctuations, and the lithium battery supplements long-term power deficits. The two work together to meet the compensation needs of disturbances of different durations, and their total output power follows the formula P. ess =P bat +P scThis ensures rapid power balancing during disturbances. For disturbances such as slow photovoltaic output decay on a minute-by-minute basis and gradual load increases, long-term power support is achieved through flexible adjustments of controllable distributed power sources (micro gas turbines, diesel generators). This hierarchical response mechanism increases the system's response speed to various disturbances to 3-5 times that of traditional methods. Within 3 seconds of a disturbance, the frequency deviation can be controlled within ±0.2Hz (grid-connected mode) or ±0.5Hz (off-grid mode), and within 10 seconds, it can recover to within ±0.1Hz (grid-connected mode) or ±0.2Hz (off-grid mode), far superior to the 15-20 second recovery time of traditional scheduling methods, greatly reducing the risk of power outages caused by disturbances.

[0156] On the other hand, the perturbation observer and dynamic correction strategy introduced in the method further enhance the accuracy of perturbation resistance. The perturbation observer is based on the extended Kalman filter algorithm, and is determined by the formula ΔP. re =P re,act -P re,pre Real-time quantification of renewable energy power forecasting errors to detect disturbance trends in advance; when the disturbance amplitude exceeds |ΔP|≥α×P rated When the preset threshold (α is 5%-15%) is reached, a fast response mechanism is automatically triggered, prioritizing the use of energy storage units for compensation, while simultaneously adjusting the controllable power output plan to prevent system instability caused by accumulated disturbances. Actual test data shows that after adopting this method, the power fluctuation variance at the microgrid grid connection point (as shown by the formula) is significantly reduced. The computation is reduced by 42% compared to traditional methods, resulting in a significant improvement in system stability.

[0157] Significantly reducing the operating costs and improving the economic efficiency of hybrid energy microgrids: High operating costs are one of the key factors restricting the large-scale application of hybrid energy microgrids. Traditional dispatching methods, due to a lack of precise design for multi-device collaborative optimization, often suffer from problems such as frequent start-ups and shutdowns of controllable power sources, overcharging and discharging of energy storage, and curtailment of renewable energy, leading to increased fuel consumption, aggravated equipment wear, and poor economic efficiency. This method achieves a significant reduction in operating costs through the design of a multi-objective optimization objective function and a refined cost control strategy.

[0158] From the perspective of objective function construction, the method takes system operating cost (F2) as one of the core optimization objectives. The priority of cost optimization is clearly defined using the formula minF=ω1F1+ω2F2+ω3F3 (ω2 has a weight of 0.5-0.6 in typical scenarios). Simultaneously, the cost components are finely broken down—including not only the fuel cost of the controllable power source but also maintenance costs and energy storage degradation costs. The fuel cost is determined by the formula... Precise calculations, combined with the cost characteristics of different controllable power sources (such as the nonlinear cost of micro gas turbines and the linear cost of diesel generators), optimize power output allocation and avoid ineffective operation of high-cost equipment. For example, during off-peak hours, low-cost renewable energy and energy storage discharge are prioritized, reducing the number of times micro gas turbines are started; during peak hours, based on differences in fuel cost coefficients, the more efficient micro gas turbines undertake the main peak-shaving tasks, while diesel generators are only used as emergency backups, effectively reducing fuel consumption.

[0159] From the perspective of equipment loss control, the method extends equipment lifespan and indirectly reduces costs through strict constraints and strategy optimization. For energy storage units, the SOC formula is used... min ≤SOC t ≤SOC max (SOC min =20%, SOC max =90%) avoids overcharging and over-discharging. Simultaneously, by dividing the workload between supercapacitors (handling high-frequency fluctuations) and lithium batteries (handling low-frequency fluctuations), the number of charge-discharge cycles for lithium batteries is reduced, extending their lifespan by 20%-30%. For controllable power sources, a smooth output adjustment strategy (with real-time adjustment correction not exceeding ±20%) avoids mechanical losses caused by frequent start-stop cycles and large output fluctuations, reducing maintenance costs by 15%-20%. In practical applications, the daily operating cost of microgrids using this method is 15% lower than traditional methods, resulting in annual cost savings of hundreds of thousands of yuan, significantly improving the economic feasibility of microgrids.

[0160] Improving the renewable energy absorption rate and promoting energy efficiency and low-carbon development: With the advancement of the "dual carbon" goals, the proportion of renewable energy in hybrid energy microgrids is constantly increasing. However, due to the strong fluctuations in their output and the difficulty in prediction, traditional dispatching methods often fail to balance power in a timely manner, resulting in a large amount of wind and solar curtailment. The renewable energy absorption rate is generally low (less than 80% in some microgrids), leading to serious energy waste. This method, through multi-dimensional optimization design, significantly improves the renewable energy absorption capacity and promotes the upgrading of energy utilization efficiency.

[0161] At the goal-oriented level, the method incorporates the energy utilization rate (F3) as a core optimization objective, dynamically adjusting its weights using the formula minF=ω1F1+ω2F2+ω3F3. During periods of high renewable energy generation (such as strong midday sunlight and abundant winds at night), ω3 is increased to above 0.5 to prioritize renewable energy consumption and avoid power curtailment. Simultaneously, F3 is calculated based on "renewable energy consumption rate + energy storage cycle efficiency," focusing on both the actual utilization of renewable energy and the energy conversion efficiency of energy storage devices, forming a closed loop of "multi-stage synergistic efficiency improvement."

[0162] At the technical strategy level, the method provides a guarantee for renewable energy consumption through a coordinated "prediction-storage-scheduling" mechanism. On the one hand, the disturbance observer accurately predicts fluctuations in renewable energy output based on real-time data and historical patterns (prediction error controlled within 5%), and formulates consumption plans in advance. On the other hand, the hybrid energy storage system acts as a "buffer," storing excess energy through charging when renewable energy output is excessive (e.g., charging the SOC from 30% to 80% during a midday solar power surge), and releasing energy through discharging when output is insufficient, avoiding power curtailment due to excess power. In addition, the improved predictive control algorithm adjusts the operating status of controllable power sources and energy storage in real time through rolling optimization, ensuring that renewable energy output can always be efficiently absorbed by the load or energy storage. For example, when solar power output suddenly increases by 100kW, the system first calls on the supercapacitor for rapid charging, then adjusts the lithium battery charging power, and appropriately reduces the output of the micro gas turbine, achieving power balance while 100% consumption of solar power. Actual tests show that after adopting this method, the renewable energy absorption rate of hybrid energy microgrids increases by more than 8%, and can reach more than 95% in some scenarios. The amount of abandoned electricity reduced each year can meet the daily electricity needs of hundreds of households, significantly reducing energy waste and promoting the transformation of microgrids towards low-carbon and high-efficiency.

[0163] Enhancing the scenario adaptability and flexible scalability of hybrid energy microgrids: Hybrid energy microgrids of different sizes and operating modes (grid-connected / off-grid) have significantly different operational needs and constraints (e.g., small microgrids have weak disturbance immunity, while medium and large microgrids are highly cost-sensitive; grid-connected modes need to focus on interaction with the main grid, while off-grid modes need to ensure independent power supply). Traditional dispatching methods often suffer from poor scalability due to their "one-size-fits-all" design, making them difficult to adapt to diverse scenarios. This method, through modular design and adaptive parameter adjustment, possesses strong scenario adaptability and can meet the operational needs of different types of microgrids.

[0164] Regarding scale adaptation, the method dynamically adjusts the threshold coefficient (α) to suit the disturbance rejection requirements of microgrids with different capacities: for small microgrids with a capacity less than 500 kVA (such as rural microgrids), the system inertia is small and the disturbance rejection capability is weak, so α is set to 10%-15% to improve response sensitivity and quickly suppress small disturbances; for medium and large microgrids with a capacity greater than 1000 kVA (such as industrial park microgrids), the system inertia is large and the load is complex, so α is set to 5%-10% to avoid increased losses caused by frequent responses. Meanwhile, in the algorithm solution stage, the population size and number of iterations of the Particle Swarm Optimization (PSO) algorithm can be flexibly adjusted according to the complexity of the microgrid (population size 30, iterations 50 for small microgrids, population size 50, iterations 100 for medium and large microgrids), balancing computational efficiency and optimization accuracy.

[0165] Regarding operational mode adaptation, the method fully considers the differences between grid-connected and off-grid modes in the construction of the objective function and the design of constraints: In grid-connected mode, the focus is on optimizing the power fluctuation at the grid connection point (through F1 quantization) to reduce the impact on the main grid; in off-grid mode, priority is given to ensuring frequency and voltage stability (by adding a frequency / voltage deviation penalty term in F1) to ensure the reliability of independent power supply. In addition, the scheduling strategy is issued using a collaborative architecture of "edge computing + cloud computing". Commands with high real-time requirements (such as energy storage control) are processed locally on edge nodes (latency ≤100ms) to adapt to the independent operation requirements in off-grid mode; global optimization commands (such as intraday plans) are computed in the cloud to adapt to the collaborative scheduling requirements with the main grid in grid-connected mode.

[0166] In terms of scalability, the method's system model adopts a modular design. When adding new power generation units (such as hydrogen fuel cells) or energy storage devices (such as vanadium redox flow batteries), only the corresponding physical characteristic parameters (such as power-efficiency curves and charge-discharge rates) and cost coefficients need to be supplemented, without reconstructing the overall algorithm framework, significantly reducing the difficulty and cost of technology upgrades. This strong adaptability and scalability allow this method to be widely applied to different scenarios such as rural microgrids, industrial park microgrids, and island microgrids, providing technical support for the large-scale promotion of hybrid energy microgrids.

[0167] Ensuring the safe operation of hybrid energy microgrids and reducing O&M risks: Hybrid energy microgrids contain multiple types of power generation, energy storage, and load equipment. These devices are closely coupled, and any safety hazard in any link can trigger a cascading failure. Traditional dispatching methods, lacking comprehensive safety constraints and real-time monitoring, are prone to equipment exceeding limits (such as overcharging of energy storage or line overload), increasing O&M risks. This method, through comprehensive safety constraint design and real-time evaluation mechanisms, constructs a "prevention-control-optimization" safety assurance system to ensure the safe and stable operation of the system.

[0168] At the preventative level, the method establishes safety constraints covering all equipment: for power generation units, it clearly defines upper and lower output limits (e.g., minimum output of 30% of rated power and maximum output of 100% of rated power for micro gas turbines) to avoid no-load or overload operation; for energy storage units, it uses SOC (State of Charge) to... min ≤SOC t ≤SOC max The system constrains the state of charge to prevent battery bulging and lifespan degradation caused by overcharging and over-discharging. For power grid lines, it sets a transmission power limit (not exceeding the thermal stability limit) to avoid line overheating and burnout. For the system as a whole, it strictly adheres to power balance constraints (total output = total load + network loss) to prevent frequency collapse caused by power imbalance. These constraints take effect in real time during the algorithm's solution process, ensuring that every scheduling instruction remains within safe boundaries.

[0169] At the in-process control level, the method mitigates safety risks in a timely manner through dynamic correction and rapid response mechanisms. When equipment operating parameters are detected to be close to limits (such as SOC approaching the lower limit of 20%), the system automatically adjusts the scheduling strategy (such as reducing energy storage discharge and increasing controllable power output) to avoid parameter over-limits. When equipment failure occurs (such as inverter overcurrent), the rapid response mechanism immediately disconnects the faulty equipment and simultaneously calls on backup equipment (such as diesel generators) to supplement power, preventing the fault from spreading.

[0170] At the post-event optimization level, the scheduling effect evaluation step of the method compares key indicators before and after the disturbance (such as frequency deviation Δf = |f) t -f nominal The system monitors the system's safety status in real time, including voltage fluctuation range. When evaluation indicators fail to meet preset standards (such as frequency deviation exceeding ±0.2Hz), it automatically initiates a parameter optimization process, adjusting algorithm parameters (such as prediction window length and weighting coefficients) and optimizing scheduling strategies to fundamentally reduce safety risks. This full-cycle safety assurance system reduces equipment failure rates in hybrid energy microgrids by more than 30%, reduces the workload of maintenance personnel by 40%, and significantly improves the safety and reliability of system operation.

[0171] In summary, the dynamic anti-disturbance coordinated scheduling method for hybrid energy microgrids described in this invention addresses key pain points in the operation of hybrid energy microgrids from five core dimensions: stability, economy, energy efficiency, scenario adaptability, and security. Through innovative technical design and refined strategy optimization, it provides strong technical support for the large-scale and high-quality application of hybrid energy microgrids, and has extremely high practical value and promotion significance.

[0172] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A method for dynamic anti-disturbance coordinated scheduling of hybrid energy microgrid, comprising the following steps: (1) constructing a system model of the hybrid energy microgrid, the system model comprising at least a renewable energy generation unit, an energy storage unit, a controllable distributed power source and a load unit, wherein each unit is connected to a common DC bus or AC bus through a power electronic converter to form a multi-source coordinated power supply network; the system model synchronously comprises physical characteristic parameters, operation constraint conditions and interface characteristics of each unit, wherein the physical characteristic parameters comprise a power-efficiency curve of the generation unit, a charge-discharge rate characteristic of the energy storage unit and a time sequence variation law of the load unit; (2) collecting real-time operation data of the hybrid energy microgrid, the real-time operation data comprising output power of each generation unit, state of charge of the energy storage unit, load demand and system disturbance information, the data collection being achieved by deploying intelligent sensing terminals at key nodes, and the time stamp synchronization technology being used to ensure spatiotemporal consistency of multi-source data, and the collected original data being subjected to filtering, outlier rejection and normalization pretreatment to form a standardized data set meeting input requirements of the scheduling algorithm; (3) establishing a dynamic anti-disturbance scheduling objective function based on the real-time operation data, the objective function aiming to suppress power fluctuation, minimize operation cost and improve energy utilization rate, and the objective function being constructed by considering differences in operation modes (grid-connected / off-grid) of the microgrid, the grid-connected mode being focused on optimizing interactive power characteristics with the main grid, and the off-grid mode being focused on guaranteeing system frequency and voltage stability; (4) solving the objective function by using an improved predictive control algorithm to obtain initial scheduling instructions of each controllable unit, the solving process comprising a constraint condition checking link, and the initial scheduling instructions being ensured to meet system safe operation boundaries by using a feasibility judgment mechanism, and an infeasible solution being modified by using a slack variable adjustment method; (5) dynamically modifying the initial scheduling instructions according to system disturbance information to generate a final coordinated scheduling strategy and issue the final coordinated scheduling strategy to each execution unit, and the dynamic modification process using a hierarchical response mechanism, millisecond-level disturbance being quickly responded by local hardware protection, second-level disturbance being compensated in real time by the energy storage unit, and minute-level disturbance being adjusted by the controllable power source to achieve accurate suppression of disturbances of different time scales. 2.The method according to claim 1, wherein the renewable energy generation unit in step (1) comprises a photovoltaic power generation module and a wind power generation module, the energy storage unit comprises a hybrid energy storage system composed of a lithium battery and a super capacitor, and the controllable distributed power source comprises a micro gas turbine and a diesel generator; and the total output power of the hybrid energy storage system satisfies: Ptotal=Plithium+Psuper capacitor. 3.The method according to claim 1, wherein the hybrid energy microgrid comprises a plurality of renewable energy generation units, a plurality of energy storage units, a plurality of controllable distributed power sources and a plurality of load units, and the system model comprises a plurality of physical characteristic parameters, operation constraint conditions and interface characteristics of each unit. ​ P ess = P bat + P sc Among them, P ess P represents the total output power of the hybrid energy storage system. bat For the output power of the lithium battery, P sc The formula represents the output power of the supercapacitor. It is used to characterize the power distribution relationship between the two energy storage devices in a hybrid energy storage system, and to achieve coordinated control in which the supercapacitor responds to short-term high-frequency disturbances and the lithium battery compensates for long-term low-frequency fluctuations. The supercapacitor is responsible for responding to power fluctuations within 10 seconds, while the lithium battery is responsible for handling power deficits of 10 seconds to 10 minutes. This division of labor can reduce the number of charge-discharge cycles of the lithium battery and extend its service life. ​ The system disturbance information in step (2) includes renewable energy power prediction error, load mutation signal, communication network delay and equipment failure information, the real-time operation data is collected through a distributed sensing network, and the sampling frequency is not less than 1 Hz; the renewable energy power prediction error calculation formula is: ΔP re = P re,act - P re,pre where ΔP re is the renewable energy power prediction error, P re,act is the actual output power, P re,pre is the predicted output power; the formula is used to quantify the uncertainty of renewable energy output, to provide disturbance amplitude judgment basis for subsequent anti-disturbance scheduling. When the prediction error is positive, it means that the actual output is higher than the predicted value, and the system needs to balance through energy storage charging or reducing renewable energy output. When the prediction error is negative, it means that the actual output is lower than the predicted value, and the system needs to call backup power or release energy storage for compensation.

4. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 1, characterized in that: The target function in step (3) is constructed in a weighted summation form, and three types of indexes, power fluctuation suppression, operation cost and energy utilization rate, are comprehensively considered, and the expression is: minF = ω1F1 + ω2F2 + ω3F3 Wherein, F1 is the power fluctuation suppression index, F2 is the system operation cost, F3 is the energy utilization rate index, ω1, ω2, ω3 are weight coefficients and satisfy ω1 + ω2 + ω3 = 1; the formula realizes the collaborative decision of multi-objective optimization through weight distribution, and the priority of each index can be dynamically adjusted according to the microgrid operation scene, in the emergency scene such as power grid fault, the ω1 weight is increased to 0.6 or more to preferentially ensure system stability, in the conventional economic operation scene, the ω2 weight is set to 0.5-0.6 to highlight the cost optimization goal, and in the renewable energy large generation period, the ω3 weight is increased to 0.5 or more to improve the consumption rate.

5. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 4, characterized in that: The power fluctuation suppression index F1 adopts microgrid grid-connected point power fluctuation variance calculation, and the expression is: where P grid,i is the grid-connected point power at the i th moment, is the average grid-connected point power, and n is the number of sampling points; this formula is used to quantify the grid-connected point power fluctuation degree, and the smaller the value, the better the system power stability. In actual calculation, the value of n is determined according to the scheduling period, n = 96 (15-minute interval, 24 hours) is used for the intra-day scheduling layer, and n = 60 (1-minute interval, 1 hour) is used for the real-time adjustment layer, and multi-scale fluctuation evaluation is realized through calculation of different window lengths.

6. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 4, characterized in that: The system operation cost F2 includes fuel cost, maintenance cost and energy storage attenuation cost, wherein the fuel cost calculation formula is: where a k , b k , c k are the fuel cost coefficients of the kth controllable power source, P k,t is the output at time t, m is the total number of controllable power sources, and Δt is the scheduling time interval; this formula is used to accurately calculate the fuel consumption cost of controllable power sources, providing a quantitative basis for economic optimization, wherein the coefficients a k , b k , c k are determined according to the type of equipment, the quadratic coefficient b k of the micro gas turbine has a large absolute value, reflecting its nonlinear cost characteristics, and the linear coefficient a k of the diesel generator dominates, reflecting its approximate linear fuel consumption law.

7. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 1, characterized in that: In step (4), the improved predictive control algorithm is based on a model predictive control framework, a disturbance observer is introduced to realize real-time prediction of renewable energy output fluctuation, and a rolling optimization window is adopted to realize multi-time scale coordination, the optimization window includes an intraday scheduling layer (15-minute level) and a real-time adjustment layer (1-minute level); the coordination relationship of the two time scales satisfies: T day = N x T real where T day is the day-ahead scheduling period (15 minutes), T real is the real-time adjustment period (1 minute), and N is an integer and N = 15; the formula is used to define the time scale correlation of different scheduling levels, to realize the organic combination of global optimization and real-time response, the day-ahead scheduling layer generates the output plan curve within 24 hours according to the ultra-short-term prediction data, and the real-time adjustment layer corrects the day-ahead plan based on the latest measurement data in each 1-minute period, and the correction amplitude does not exceed ±20% of the day-ahead plan value, to ensure the smoothness of the scheduling strategy.

8. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 1, characterized in that: In step (5), the dynamic correction process includes: when it is detected that the system disturbance amplitude exceeds a preset threshold, a fast response mechanism is triggered, the power compensation is preferentially called by the energy storage unit, and the output plan of the controllable distributed power supply is adjusted, the preset threshold is set to 5%-15% of the system rated power according to the microgrid capacity level; the disturbance amplitude judgment formula is: |ΔP| ≥ α x P rated Wherein, |ΔP| is the absolute value of the disturbance power, α is the threshold coefficient (5%-15%), P rated is the system rated power; the formula is used to determine whether to trigger the fast response mechanism, to ensure that the disturbance is effectively suppressed before exceeding the system tolerance range. For small microgrids with a capacity of less than 500 kVA, the value of α is 10%-15% to improve the response sensitivity. For medium and large microgrids with a capacity of more than 1000 kVA, the value of α is 5%-10% to avoid the increase in operating loss caused by frequent response.

9. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 1, characterized in that: Further comprising establishing a safety constraint condition, the safety constraint condition includes: upper and lower limit constraints of each power generation unit output, energy storage unit state of charge constraints (20%-90%), system power balance constraints and network power flow constraints; wherein the energy storage unit state of charge constraint expression is: SOC min ≤ SOC t ≤ SOC max Wherein, SOC t is the state of charge at time t, SOC min = 20%, SOC max = 90%; this formula is used to prevent overcharging and over-discharging of the energy storage unit, prolong the service life of the device and ensure the scheduling flexibility. In actual operation, when it is predicted that there will be a large renewable energy generation period in the next 24 hours, the SOC will be adjusted to below 30% in advance to reserve charging space. When it is predicted that there will be a load peak or severe weather leading to a sharp decrease in renewable energy output, the SOC will be raised to above 80% in advance to reserve emergency power.

10. The hybrid energy microgrid dynamic anti-disturbance coordinated scheduling method according to claim 1, characterized in that: It further comprises a scheduling effect evaluation step: by comparing the system frequency deviation, voltage fluctuation range and economic index changes before and after the disturbance occurs, the effectiveness of the dynamic anti-disturbance coordinated scheduling method is evaluated, and when the evaluation index does not meet the preset requirement, the algorithm parameters are automatically updated and the scheduling process is re-executed; the system frequency deviation calculation formula is: Δf = |f t -f nominal | Wherein, Δf is the absolute value of frequency deviation, f t is the system frequency at time t, f nominal is the rated frequency (50 Hz or 60 Hz); the formula is used to quantify the system frequency stability, and is one of the core indicators for evaluating the anti-disturbance effect. The evaluation standard is that within 3 seconds after the disturbance occurs, Δf needs to be controlled within ±0.2 Hz (grid-connected mode) or ±0.5 Hz (off-grid mode), and within 10 seconds, it is restored to within ±0.1 Hz. When the evaluation does not meet the standard for three consecutive times, the algorithm parameter optimization process is automatically started, and the prediction control window length and weight coefficient are readjusted.

Images